fill tag value for sample and add sample test cases

src/client/src/tscSQLParser.c

@@ -2830,15 +2830,16 @@ int32_t addExprAndResultField(SSqlCmd* pCmd, SQueryInfo* pQueryInfo, int32_t col
 
         // todo REFACTOR
         // set the first column ts for top/bottom query
+        int32_t tsFuncId = (functionId == TSDB_FUNC_MAVG) ? TSDB_FUNC_TS_DUMMY : TSDB_FUNC_TS;
         SColumnIndex index1 = {index.tableIndex, PRIMARYKEY_TIMESTAMP_COL_INDEX};
-        pExpr = tscExprAppend(pQueryInfo, TSDB_FUNC_TS_DUMMY, &index1, TSDB_DATA_TYPE_TIMESTAMP, TSDB_KEYSIZE, getNewResColId(pCmd),
+        pExpr = tscExprAppend(pQueryInfo, tsFuncId, &index1, TSDB_DATA_TYPE_TIMESTAMP, TSDB_KEYSIZE, getNewResColId(pCmd),
                               0, false);
-        tstrncpy(pExpr->base.aliasName, aAggs[TSDB_FUNC_TS_DUMMY].name, sizeof(pExpr->base.aliasName));
+        tstrncpy(pExpr->base.aliasName, aAggs[tsFuncId].name, sizeof(pExpr->base.aliasName));
 
         const int32_t TS_COLUMN_INDEX = PRIMARYKEY_TIMESTAMP_COL_INDEX;
         SColumnList   ids = createColumnList(1, index.tableIndex, TS_COLUMN_INDEX);
         insertResultField(pQueryInfo, colIndex, &ids, TSDB_KEYSIZE, TSDB_DATA_TYPE_TIMESTAMP,
-                          aAggs[TSDB_FUNC_TS_DUMMY].name, pExpr);
+                          aAggs[tsFuncId].name, pExpr);
 
         colIndex += 1;  // the first column is ts
 
@@ -5585,7 +5586,8 @@ int32_t validateFillNode(SSqlCmd* pCmd, SQueryInfo* pQueryInfo, SSqlNode* pSqlNo
   size_t numOfExprs = tscNumOfExprs(pQueryInfo);
   for(int32_t i = 0; i < numOfExprs; ++i) {
     SExprInfo* pExpr = tscExprGet(pQueryInfo, i);
-    if (pExpr->base.functionId == TSDB_FUNC_TOP || pExpr->base.functionId == TSDB_FUNC_BOTTOM) {
+    if (pExpr->base.functionId == TSDB_FUNC_TOP || pExpr->base.functionId == TSDB_FUNC_BOTTOM
+        || pExpr->base.functionId == TSDB_FUNC_SAMPLE) {
       return invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg3);
     }
   }

src/query/src/qAggMain.c

@@ -186,6 +186,7 @@ typedef struct  {
   int16_t colBytes;
   char *values;
   int64_t *timeStamps;
+  char  *taglists;
 } SSampleFuncInfo;
 
 int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionId, int32_t param, int16_t *type,
@@ -323,7 +324,7 @@ int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionI
       return TSDB_CODE_SUCCESS;
     } else if (functionId == TSDB_FUNC_SAMPLE) {
       *type = TSDB_DATA_TYPE_BINARY;
-      *bytes = (int16_t)(sizeof(SSampleFuncInfo) + dataBytes *  param + sizeof(int64_t) * param);
+      *bytes = (int16_t)(sizeof(SSampleFuncInfo) + dataBytes*param + sizeof(int64_t)*param + extLength*param);
       *interBytes = *bytes;
 
       return TSDB_CODE_SUCCESS;
@@ -438,7 +439,7 @@ int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionI
   } else if (functionId == TSDB_FUNC_SAMPLE) {
       *type = (int16_t)dataType;
       *bytes = (int16_t)dataBytes;
-      size_t size = sizeof(SSampleFuncInfo) + sizeof(dataBytes) *  param + sizeof(int64_t) * param;
+      size_t size = sizeof(SSampleFuncInfo) + dataBytes*param + sizeof(int64_t)*param + extLength*param;
       *interBytes = (int32_t)size;
   } else if (functionId == TSDB_FUNC_LAST_ROW) {
     *type = (int16_t)dataType;
@@ -4684,21 +4685,38 @@ static SSampleFuncInfo* getSampleFuncOutputInfo(SQLFunctionCtx *pCtx) {
   }
 }
 
-static void assignResultSample(SSampleFuncInfo *pInfo, int32_t index, int64_t ts, void *pData, uint16_t type, int16_t bytes) {
+static void assignResultSample(SQLFunctionCtx *pCtx, SSampleFuncInfo *pInfo, int32_t index, int64_t ts, void *pData, uint16_t type, int16_t bytes, char *inputTags) {
   assignVal(pInfo->values + index*bytes, pData, bytes, type);
   *(pInfo->timeStamps + index) = ts;
-  return;
+
+  SExtTagsInfo* pTagInfo = &pCtx->tagInfo;
+  int32_t posTag = 0;
+  char* tags = pInfo->taglists + pTagInfo->tagsLen;
+  if (pCtx->currentStage == MERGE_STAGE) {
+    memcpy(tags, inputTags, (size_t)pTagInfo->tagsLen);
+  } else {
+    for (int32_t i = 0; i < pTagInfo->numOfTagCols; ++i) {
+      SQLFunctionCtx* ctx = pTagInfo->pTagCtxList[i];
+      if (ctx->functionId == TSDB_FUNC_TS_DUMMY) {
+        ctx->tag.nType = TSDB_DATA_TYPE_BIGINT;
+        ctx->tag.i64 = ts;
+      }
+
+      tVariantDump(&ctx->tag, tags + posTag, ctx->tag.nType, true);
+      posTag += pTagInfo->pTagCtxList[i]->outputBytes;
+    }
+  }
 }
 
-static void do_reservoir_sample(SSampleFuncInfo *pInfo, int32_t samplesK, int64_t ts, void *pData,  uint16_t type, int16_t bytes) {
+static void do_reservoir_sample(SQLFunctionCtx *pCtx, SSampleFuncInfo *pInfo, int32_t samplesK, int64_t ts, void *pData,  uint16_t type, int16_t bytes) {
   pInfo->totalPoints++;
   if (pInfo->numSampled < samplesK) {
-    assignResultSample(pInfo, pInfo->numSampled, ts, pData, type, bytes);
+    assignResultSample(pCtx, pInfo, pInfo->numSampled, ts, pData, type, bytes, NULL);
     pInfo->numSampled++;
   } else {
     int32_t j = rand() % (pInfo->totalPoints);
     if (j < samplesK) {
-      assignResultSample(pInfo, j, ts, pData, type, bytes);
+      assignResultSample(pCtx, pInfo, j, ts, pData, type, bytes, NULL);
     }
   }
 }
@@ -4712,11 +4730,25 @@ static void copySampleFuncRes(SQLFunctionCtx *pCtx, int32_t type) {
   for (int32_t i = 0; i < pRes->numSampled; ++i) {
     assignVal(pOutput, pRes->values + i*pRes->colBytes, pRes->colBytes, type);
     *pTimestamp = *(pRes->timeStamps + i);
-
     pOutput += pCtx->outputBytes;
     pTimestamp++;
   }
 
+  char **pData = calloc(pCtx->tagInfo.numOfTagCols, POINTER_BYTES);
+  for (int32_t i = 0; i < pCtx->tagInfo.numOfTagCols; ++i) {
+    pData[i] = pCtx->tagInfo.pTagCtxList[i]->pOutput;
+  }
+
+  for (int32_t i = 0; i < pRes->numSampled; ++i) {
+    int16_t tagOffset = 0;
+    for (int32_t j = 0; j < pCtx->tagInfo.numOfTagCols; ++j) {
+      memcpy(pData[j], pRes->taglists + i*pCtx->tagInfo.tagsLen + tagOffset, (size_t)pCtx->tagInfo.pTagCtxList[j]->outputBytes);
+      tagOffset += pCtx->tagInfo.pTagCtxList[j]->outputBytes;
+      pData[j] += pCtx->tagInfo.pTagCtxList[j]->outputBytes;
+    }
+  }
+
+  tfree(pData);
 }
 
 static bool sample_function_setup(SQLFunctionCtx *pCtx, SResultRowCellInfo* pResInfo) {
@@ -4732,6 +4764,7 @@ static bool sample_function_setup(SQLFunctionCtx *pCtx, SResultRowCellInfo* pRes
   pRes->values = ((char*)pRes + sizeof(SSampleFuncInfo));
   pRes->colBytes = (pCtx->currentStage != MERGE_STAGE) ? pCtx->inputBytes : pCtx->outputBytes;
   pRes->timeStamps = (int64_t *)((char *)pRes->values + pRes->colBytes * pCtx->param[0].i64);
+  pRes->taglists = (char*)pRes->timeStamps + sizeof(int64_t) * pCtx->param[0].i64;
   return true;
 }
 
@@ -4744,6 +4777,7 @@ static void sample_function(SQLFunctionCtx *pCtx) {
   if (pRes->values !=  ((char*)pRes + sizeof(SSampleFuncInfo))) {
     pRes->values =  ((char*)pRes + sizeof(SSampleFuncInfo));
     pRes->timeStamps = (int64_t*)((char*)pRes->values + pRes->colBytes * pCtx->param[0].i64);
+    pRes->taglists = (char*)pRes->timeStamps + sizeof(int64_t) * pCtx->param[0].i64;
   }
 
   for (int32_t i = 0; i < pCtx->size; ++i) {
@@ -4755,7 +4789,7 @@ static void sample_function(SQLFunctionCtx *pCtx) {
     notNullElems++;
 
     TSKEY ts = (pCtx->ptsList != NULL)? GET_TS_DATA(pCtx, i):0;
-    do_reservoir_sample(pRes, (int32_t)pCtx->param[0].i64, ts, data, pCtx->inputType, pRes->colBytes);
+    do_reservoir_sample(pCtx, pRes, (int32_t)pCtx->param[0].i64, ts, data, pCtx->inputType, pRes->colBytes);
   }
 
   if (!pCtx->hasNull) {
@@ -4774,13 +4808,15 @@ static void sample_func_merge(SQLFunctionCtx *pCtx) {
   SSampleFuncInfo* pInput = (SSampleFuncInfo*)GET_INPUT_DATA_LIST(pCtx);
   pInput->values = ((char*)pInput + sizeof(SSampleFuncInfo));
   pInput->timeStamps = (int64_t*)((char*)pInput->values + pInput->colBytes * pCtx->param[0].i64);
+  pInput->taglists = (char*)pInput->timeStamps + sizeof(int64_t)*pCtx->param[0].i64;
 
   SSampleFuncInfo *pOutput = getSampleFuncOutputInfo(pCtx);
   pOutput->totalPoints = pInput->totalPoints;
   pOutput->numSampled = pInput->numSampled;
   for (int32_t i = 0; i < pInput->numSampled; ++i) {
-    assignResultSample(pOutput, i, pInput->timeStamps[i],
-                       pInput->values + i * pInput->colBytes, pCtx->outputType, pInput->colBytes);
+    assignResultSample(pCtx, pOutput, i, pInput->timeStamps[i],
+                       pInput->values + i * pInput->colBytes, pCtx->outputType, pInput->colBytes,
+                       pInput->taglists + i*pCtx->tagInfo.tagsLen);
   }
 
   SET_VAL(pCtx, pInput->numSampled, pOutput->numSampled);

src/query/src/qExecutor.c

@@ -3986,7 +3986,9 @@ void setResultOutputBuf(SQueryRuntimeEnv *pRuntimeEnv, SResultRow *pResult, SQLF
     offset += pCtx[i].outputBytes;
 
     int32_t functionId = pCtx[i].functionId;
-    if (functionId == TSDB_FUNC_TOP || functionId == TSDB_FUNC_BOTTOM || functionId == TSDB_FUNC_DIFF || functionId == TSDB_FUNC_DERIVATIVE) {
+    if (functionId == TSDB_FUNC_TOP || functionId == TSDB_FUNC_BOTTOM ||
+        functionId == TSDB_FUNC_DIFF || functionId == TSDB_FUNC_DERIVATIVE ||
+        functionId == TSDB_FUNC_SAMPLE || functionId == TSDB_FUNC_MAVG || functionId == TSDB_FUNC_CSUM) {
       if(i > 0) pCtx[i].ptsOutputBuf = pCtx[i-1].pOutput;
     }
 

tests/script/general/parser/col_arithmetic_operation.sim

@@ -124,6 +124,7 @@ sql select spread(ts )/(1000*3600*24) from $stb interval(1y)
 sql_error select first(c1, c2) - last(c1, c2) from $stb interval(1y)
 sql_error select first(ts) - last(ts) from $stb interval(1y)
 sql_error select top(c1, 2) - last(c1) from $stb;
+sql_error select sample(c1, 2) - last(c1) from $stb;
 sql_error select stddev(c1) - last(c1) from $stb;
 sql_error select diff(c1) - last(c1) from $stb;
 sql_error select csum(c1) - last(c1) from $stb;

tests/script/general/parser/col_arithmetic_query.sim

@@ -174,6 +174,9 @@ endi
 sql_error select top(c1, 1) - bottom(c1, 1) from $tb
 sql_error select top(c1, 99) - bottom(c1, 99) from $tb
 sql_error select top(c1,1) - 88 from $tb
+sql_error select sample(c1, 1) - bottom(c1, 1) from $tb
+sql_error select sample(c1, 99) - bottom(c1, 99) from $tb
+sql_error select sample(c1,1) - 88 from $tb
 
 # all data types [d.6] ================================================================
 sql select c2-c1*1.1, c3/c2, c4*c3, c5%c4, (c6+c4)%22, c2-c2 from $tb
@@ -475,11 +478,16 @@ endi
 sql_error select first(c1, c2) - last(c1, c2) from $stb
 sql_error select top(c1, 5) - bottom(c1, 5) from $stb
 sql_error select first(*) - 99 from $stb
+sql_error select sample(c1, 5) - bottom(c1, 5) from $stb
+
 
 # multi row result aggregation [d.4]
 sql_error select top(c1, 1) - bottom(c1, 1) from $stb
 sql_error select top(c1, 99) - bottom(c1, 99) from $stb
 
+sql_error select sample(c1, 1) - top(c1, 1) from $stb
+sql_error select sample(c1, 99) - top(c1, 99) from $stb
+
 # query on super table [d.5]=============================================================
 # all cases in this part are query on super table
 

tests/script/general/parser/having.sim

@@ -121,6 +121,7 @@ if $data31 != 4 then
   return -1
 endi
 
+sql_error select sample(f1,2) from st2 group by f1 having count(f2) > 0;
 sql_error select top(f1,2) from st2 group by f1 having count(f2) > 0;
 
 sql select last(f1) from st2 group by f1 having count(f2) > 0;
@@ -140,9 +141,12 @@ if $data30 != 4 then
   return -1
 endi
 
-sql_error select top(f1,2) from st2 group by f1 having count(f2) > 0;
-sql_error select top(f1,2) from st2 group by f1 having count(f2) > 0;
-sql_error select top(f1,2) from st2 group by f1 having avg(f1) > 0;
+sql_error select sample(f1,2) from st2 group by f1 having count(f2) > 0;
+sql_error select sample(f1,2) from st2 group by f1 having count(f2) > 0;
+sql_error select sample(f1,2) from st2 group by f1 having avg(f1) > 0;
+sql_error select sample(f1,2) from st2 group by f1 having count(f2) > 0;
+sql_error select sample(f1,2) from st2 group by f1 having count(f2) > 0;
+sql_error select sample(f1,2) from st2 group by f1 having avg(f1) > 0;
 
 sql select avg(f1),count(f1) from st2 group by f1 having avg(f1) > 2;
 if $rows != 2 then
@@ -1059,6 +1063,13 @@ if $data26 != 4 then
 endi  
 
 
+sql_error select avg(f1),count(st2.*),sum(f1),stddev(f1),min(f1),max(f1),first(f1),last(f1) from st2 group by f1 having sample(f1,1);
+
+sql_error select avg(f1),count(st2.*),sum(f1),stddev(f1),min(f1),max(f1),first(f1),last(f1) from st2 group by f1 having sample(f1,1) > 1;
+
+sql_error select avg(f1),count(st2.*),sum(f1),stddev(f1),min(f1),max(f1),first(f1),last(f1),sample(f1,1) from st2 group by f1 having sum(f1) > 1;
+
+sql_error select avg(f1),count(st2.*),sum(f1),stddev(f1),min(f1),max(f1),first(f1),last(f1),sample(f1,1),bottom(f1,1) from st2 group by f1 having bottom(f1,1) > 1;
 
 sql_error select avg(f1),count(st2.*),sum(f1),stddev(f1),min(f1),max(f1),first(f1),last(f1) from st2 group by f1 having top(f1,1);
 
@@ -1840,6 +1851,7 @@ if $data04 != 1 then
   return -1
 endi
 
+sql_error select sample(f1,2) from tb1 group by f1 having count(f1) > 0;
 sql_error select top(f1,2) from tb1 group by f1 having count(f1) > 0;
 sql_error select count(*) from tb1 group by f1 having last(*) > 0;
 

tests/script/general/parser/limit.sim

@@ -80,4 +80,7 @@ sql use $db
 sql select * from (select ts, top(c1, 5) from $tb where ts >= $ts0 order by ts desc limit 3 offset 1)
 sql select * from (select ts, top(c1, 5) from $stb where ts >= $ts0 order by ts desc limit 3 offset 1)
 
-system sh/exec.sh -n dnode1 -s stop -x SIGINT
+sql select * from (select ts, sample(c1, 5) from $tb where ts >= $ts0 order by ts desc limit 3 offset 1)
\ No newline at end of file
+sql_error select * from (select ts, sample(c1, 5) from $stb where ts >= $ts0 order by ts desc limit 3 offset 1)
+
+system sh/exec.sh -n dnode1 -s stop -x SIGINT

tests/script/general/parser/limit_stb.sim

@@ -828,6 +828,8 @@ if $data59 != 4 then
   return -1
 endi
 
+sql_error select sample(c1, 1) from $stb where ts >= $ts0 and ts <= $tsu limit 5 offset 1
+
 sql select top(c1, 1) from $stb where ts >= $ts0 and ts <= $tsu limit 5 offset 1
 if $rows != 0 then
   return -1

tests/script/general/parser/limit_tb.sim

@@ -355,6 +355,21 @@ sql select top(c1, 1) from $tb where ts >= $ts0 and ts <= $tsu limit 5 offset 1
 if $rows != 0 then
   return -1
 endi
+sql select sample(c1, 1) from $tb where ts >= $ts0 and ts <= $tsu limit 5 offset 1
+if $rows != 0 then
+  return -1
+endi
+
+sql select * from (select ts, sample(c1, 5) from $tb where ts >= $ts0 and ts <= $tsu order by ts desc limit 3 offset 1)
+
+sql select ts,sample(c1, 5) from $tb where ts >= $ts0 and ts <= $tsu order by ts desc limit 3 offset 1
+if $rows != 3 then
+  return -1
+endi
+print select ts,sample(c1, 5) from $tb where ts >= $ts0 and ts <= $tsu order by ts desc limit 3 offset 1
+print $data00 $data01 $data02
+print $data10 $data11 $data12
+print $data20 $data21 $data22
 
 print ========> TD-6017
 sql select * from (select ts, top(c1, 5) from $tb where ts >= $ts0 and ts <= $tsu order by ts desc limit 3 offset 1)

tests/script/general/parser/nestquery.sim

@@ -335,6 +335,8 @@ if $data12 != 71680.000000000 then
   return -1
 endi
 
+sql select sample(x, 20) from (select c1 x from nest_tb0);
+
 sql select top(x, 20) from (select c1 x from nest_tb0);
 
 sql select bottom(x, 20) from (select c1 x from nest_tb0)

tests/script/general/parser/projection_limit_offset.sim

@@ -296,6 +296,7 @@ sql_error select last(t1) from group_mt0;
 sql_error select min(t1) from group_mt0;
 sql_error select max(t1) from group_mt0;
 sql_error select top(t1, 20) from group_mt0;
+sql_error select sample(t1, 20) from group_mt0;
 sql_error select bottom(t1, 20) from group_mt0;
 sql_error select avg(t1) from group_mt0;
 sql_error select percentile(t1, 50) from group_mt0;

tests/script/general/parser/udf_dll_stable.sim

@@ -508,6 +508,7 @@ sql_error select ts,sum_double(f1),f1 from tb1;
 sql_error select add_one(f1),count(f1) from tb1;
 sql_error select sum_double(f1),count(f1) from tb1;
 sql_error select add_one(f1),top(f1,3) from tb1;
+sql_error select add_one(f1),sample(f1,3) from tb1;
 sql_error select add_one(f1) from tb1 interval(10a);